Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A method of processing at least a portion of a digital image comprising rows of pixels extending in two perpendicular directions over a 2D field, the method comprising: for each pixel in at least a row of pixels within said image portion, calculating a cumulative sum for said pixel as a sum of a value of said pixel and a value representing a sum of values of preceding pixels in the same row of pixels, wherein the values used for calculating the cumulative sum for said pixel include only values of pixels in the same row of pixels; and convolving a region of a kernel with the image portion, wherein a target pixel in the image is aligned with a specified element of the kernel region, including at least determining a difference between the cumulative sum for said pixel and a cumulative sum for at least one other pixel of the row of pixels, to determine a value for the target pixel.
This invention relates to digital image processing, specifically methods for efficiently computing convolution operations on pixel data. The problem addressed is the computational cost of traditional convolution methods, which often require repeated access to pixel values, leading to inefficiencies in processing time and memory usage. The method processes a digital image composed of rows of pixels arranged in two perpendicular directions over a 2D field. For each pixel in at least one row of the image, a cumulative sum is calculated. This sum includes the pixel's own value plus the sum of values from all preceding pixels in the same row. The calculation is restricted to pixels within the same row, ensuring locality of reference and reducing memory access overhead. The method then performs a convolution operation by aligning a target pixel in the image with a specified element of a kernel region. The convolution involves determining a difference between the cumulative sum of the target pixel and the cumulative sum of at least one other pixel in the same row. This difference is used to compute a value for the target pixel, leveraging the precomputed cumulative sums to accelerate the convolution process. The approach minimizes redundant calculations by reusing intermediate results, improving computational efficiency.
2. A method according to claim 1 , further comprising providing a plurality of pre-defined kernels for selection by a user, receiving from a user a selection of one of said pre-defined kernels, and using a region of said selected kernel as the kernel region for processing said image portion.
This invention relates to image processing, specifically methods for applying kernel-based operations to image portions. The problem addressed is the need for flexible and user-configurable kernel selection in image processing tasks, such as filtering, convolution, or feature extraction. Traditional methods often rely on fixed or manually defined kernels, which may not adapt well to varying image characteristics or user requirements. The method involves providing a plurality of pre-defined kernels, each representing a different mathematical or functional operation that can be applied to an image portion. These kernels may include common filters (e.g., Gaussian, Sobel, Laplacian) or custom-designed patterns. A user selects one of these pre-defined kernels, and a region of the selected kernel is then applied to process the image portion. This allows dynamic adaptation of the processing operation based on user input, improving flexibility and accuracy in tasks like edge detection, noise reduction, or texture analysis. The method ensures that the kernel region is properly aligned and scaled to match the image portion, ensuring consistent and meaningful results. By offering pre-defined kernels, the approach simplifies the process for users who may lack expertise in kernel design while still providing customization options. This enhances the efficiency and effectiveness of image processing workflows.
3. A method according to claim 1 , wherein the pixel value is an intensity value.
A method for processing image data involves determining a pixel value, where the pixel value is specifically an intensity value. This method is part of a broader approach for analyzing or modifying images, where pixel values are used to represent brightness or grayscale levels. The intensity value may be derived from a single channel in a grayscale image or from a luminance component in a color image. The method can be applied in various imaging applications, such as image enhancement, compression, or object detection, where accurate intensity representation is critical. By focusing on intensity values, the method ensures consistent processing across different image types and lighting conditions, improving reliability in tasks like edge detection, thresholding, or noise reduction. The technique may also involve comparing or transforming intensity values to achieve desired visual or analytical outcomes. This approach is particularly useful in fields like medical imaging, surveillance, and computer vision, where precise intensity analysis is essential for accurate results.
4. A method according to claim 1 , wherein the pixel value is a colour value.
A method for processing image data involves determining a pixel value for a pixel in an image, where the pixel value represents a color value. The method includes analyzing the pixel's neighborhood to identify a set of candidate pixels that meet specific criteria, such as spatial proximity or similarity in color. From these candidates, a subset is selected based on predefined rules, such as minimizing a cost function or maximizing a similarity metric. The pixel value is then adjusted based on the selected subset, ensuring consistency with neighboring pixels while preserving image details. This approach is particularly useful in image enhancement, noise reduction, or color correction, where maintaining accurate color representation is critical. The method may involve iterative refinement, where the selection and adjustment steps are repeated until a desired level of accuracy or convergence is achieved. The technique is applicable to various imaging applications, including digital photography, medical imaging, and computer vision, where precise color reproduction and noise reduction are essential.
5. A method according to claim 1 , wherein the portion of the digital image is a background portion of said image.
A method for processing digital images focuses on identifying and analyzing specific portions of an image, particularly background regions. The method involves segmenting the image to isolate the background portion, which is then processed to enhance or modify its visual characteristics. This approach is useful in applications where background elements need to be distinguished from foreground objects, such as in image editing, object detection, or scene reconstruction. By targeting the background portion, the method ensures that foreground elements remain unaffected, preserving their integrity while allowing adjustments to the background. The technique may involve techniques like color correction, noise reduction, or depth estimation to improve the background's appearance or usability. This method is particularly valuable in scenarios where background elements are cluttered or visually distracting, as it enables selective processing to optimize the overall image composition. The approach can be applied in various fields, including photography, video production, and computer vision, where precise control over image regions is essential.
6. A method according to claim 1 , further comprising rotating a predecessor digital image through an angle to provide said digital image.
A method for processing digital images involves rotating a predecessor digital image by a specified angle to generate a new digital image. This rotation is applied to adjust the orientation of the predecessor image, which may be necessary for alignment, correction, or further processing. The method ensures that the rotated image retains the necessary visual and structural integrity for subsequent applications, such as object recognition, pattern matching, or image analysis. The rotation process may involve interpolation techniques to maintain image quality and avoid artifacts. This technique is particularly useful in applications where image alignment or orientation correction is required, such as in medical imaging, satellite imagery, or industrial inspection systems. The method may be part of a larger image processing pipeline that includes additional steps like filtering, enhancement, or feature extraction. By rotating the predecessor image, the method enables accurate and consistent processing of digital images in various domains.
7. A method according to claim 1 , wherein the digital image is represented by Cartesian coordinates.
A method for processing digital images represented by Cartesian coordinates involves analyzing and manipulating image data in a two-dimensional coordinate system. The method addresses the challenge of efficiently handling digital images by leveraging Cartesian coordinates to simplify calculations and transformations. This approach allows for precise spatial referencing, enabling accurate image analysis, enhancement, and modification. The method may include steps such as converting image data into Cartesian coordinates, performing geometric transformations, applying filters, or extracting features based on the coordinate system. By using Cartesian coordinates, the method ensures compatibility with standard image processing algorithms and facilitates integration with other systems that rely on coordinate-based representations. The technique is particularly useful in applications requiring high precision, such as medical imaging, computer vision, and remote sensing, where accurate spatial relationships are critical. The method may also include preprocessing steps to prepare the image data for coordinate-based operations, ensuring optimal performance and accuracy in subsequent processing stages.
8. A method according to claim 1 , wherein the digital image is represented by polar coordinates.
A method for processing digital images involves representing the image in polar coordinates to enhance analysis or manipulation. The image is converted from Cartesian coordinates to polar coordinates, where each pixel is defined by a radial distance and an angular position relative to a central point. This representation simplifies certain operations, such as rotation or radial symmetry analysis, by reducing computational complexity. The method may include additional steps like filtering, feature extraction, or pattern recognition, which are optimized for the polar coordinate system. By using polar coordinates, the method improves efficiency in tasks requiring rotational invariance or radial symmetry detection, such as object recognition, medical imaging, or satellite imagery analysis. The conversion process ensures that the image data retains its integrity while enabling faster and more accurate processing. This approach is particularly useful in applications where traditional Cartesian-based methods are computationally intensive or less effective. The method can be applied to various imaging systems, including cameras, scanners, and sensors, to improve performance in real-time or high-resolution imaging tasks.
9. A method according to claim 8 , wherein said digital image represented by polar coordinates is derived from a predecessor digital image represented by Cartesian coordinates, wherein in said polar digital image, rows of pixels in one orthogonal direction correspond to respective radial distances from an origin in said Cartesian predecessor image, and rows of pixels in the other orthogonal direction correspond to respective angular displacements about said origin in said Cartesian predecessor image.
This invention relates to digital image processing, specifically converting images from Cartesian to polar coordinates for enhanced analysis or manipulation. The problem addressed is the difficulty in analyzing or processing certain image features, such as radial patterns or circular structures, when represented in standard Cartesian coordinates. Converting the image to polar coordinates simplifies such analysis by aligning these features along straight lines or uniform angular segments. The method involves transforming a digital image from Cartesian to polar coordinates. In the resulting polar image, one set of pixel rows corresponds to radial distances from a central origin point in the original Cartesian image, while the other set of pixel rows corresponds to angular displacements around that origin. This transformation allows for easier extraction of radial or angular features, such as detecting concentric patterns, analyzing rotational symmetry, or applying filters optimized for polar representations. The technique is particularly useful in applications like medical imaging, satellite imagery, or any field requiring analysis of circular or radial structures. The conversion process ensures that the polar image retains the original image's information while enabling more efficient processing of specific geometric features.
10. A method according to claim 1 further comprising setting a resolution for an image including said processed target pixel value in response to user interaction.
This invention relates to image processing, specifically adjusting image resolution based on user interaction. The method processes an image by modifying pixel values, including a target pixel value, to enhance or alter the image. The enhancement may involve noise reduction, contrast adjustment, or other image quality improvements. After processing, the method allows a user to interact with the image, such as by selecting a region or adjusting parameters, and in response, sets a resolution for the image that includes the processed target pixel value. The resolution setting may involve upscaling, downscaling, or maintaining the original resolution, depending on the user's input. The method ensures that the processed pixel values are preserved or optimized at the selected resolution. This approach provides flexibility in image resolution adjustments while maintaining the integrity of processed pixel data. The invention is useful in applications requiring dynamic resolution changes, such as digital photography, medical imaging, or real-time video processing.
11. A method according to claim 1 wherein said kernel comprises one row of contiguous elements of the same non-zero value, said method comprising setting a length of said kernel in response to user interaction.
A method for image processing involves using a kernel with a single row of contiguous elements, all having the same non-zero value. The kernel is applied to an image to perform operations such as convolution, filtering, or edge detection. The length of this kernel can be adjusted dynamically based on user input, allowing for real-time customization of the processing effect. This approach simplifies the kernel structure while providing flexibility in image analysis tasks. The method ensures that the kernel maintains a uniform value across its elements, which can be crucial for certain filtering techniques where consistency in the kernel's values is required. By allowing the user to modify the kernel length interactively, the system adapts to different image processing needs without requiring predefined kernel sizes. This technique is particularly useful in applications where the processing parameters need to be fine-tuned based on user preferences or specific image characteristics. The method streamlines the kernel design while maintaining the ability to adjust its dimensions for optimal performance.
12. A method according to claim 6 further comprising setting said angle for said rotation in response to user interaction.
A method for adjusting the angle of rotation of a component in a mechanical or electromechanical system involves dynamically modifying the rotation angle based on user input. The system includes a rotatable component, such as a motor, actuator, or mechanical linkage, and a control mechanism that governs its movement. The method further includes detecting user interaction, such as a manual adjustment or a command from a user interface, and adjusting the rotation angle of the component in response. This allows for real-time customization of the component's position or orientation, improving adaptability in applications like robotics, automation, or user-controlled devices. The method ensures precise control by translating user input into specific angular adjustments, enhancing functionality and user experience. The system may include sensors or feedback mechanisms to verify the angle setting and ensure accuracy. This approach is particularly useful in systems requiring dynamic adjustments, such as adjustable mounts, robotic arms, or interactive displays.
13. A method according to claim 1 , wherein: the row is a first row; the method further comprises, for each pixel in at least a second row of pixels within said image portion, calculating a cumulative sum for said pixel as a sum of a value of said pixel and a value representing a sum of values of preceding pixel values pixels in the second row of said image portion pixels; and convolving the region of the kernel with the image portion further includes at least determining a difference between the cumulative sum for said pixel and a cumulative sum for at least one other pixel of the second row of pixels, to determine the value for the target pixel.
This invention relates to image processing, specifically optimizing convolution operations for image analysis or computer vision tasks. The problem addressed is the computational inefficiency of traditional convolution methods, which require repeated calculations for overlapping regions of an image, leading to redundant processing and slower performance. The method involves processing an image portion by calculating cumulative sums for pixels in multiple rows. For a first row of pixels, the method computes a cumulative sum for each pixel as the sum of its value and the sum of preceding pixel values in the same row. This cumulative sum is stored for later use. For at least a second row of pixels, the method similarly calculates cumulative sums for each pixel. During convolution, the method determines the value for a target pixel by computing the difference between the cumulative sum of the target pixel and the cumulative sum of at least one other pixel in the same row. This difference represents the sum of pixel values within a defined region of a convolution kernel, enabling efficient computation without redundant calculations. The approach reduces the number of arithmetic operations needed for convolution, improving processing speed and efficiency.
14. A method according to claim 13 , wherein: convolving the region of the kernel with the image portion further includes at least determining a sum of the difference between the cumulative sums for the two pixels of the second row and the difference between the cumulative sums for the two pixels of the second row, to determine the value for the target pixel; and multiplying the differences sum by the element values of the kernel region.
This invention relates to image processing techniques, specifically methods for efficiently computing convolution operations in image analysis. The problem addressed is the computational inefficiency of traditional convolution methods, which often require redundant calculations when processing image data. The invention provides an optimized approach to convolution by leveraging cumulative sums and kernel element values to reduce the number of arithmetic operations. The method involves processing an image portion by convolving a region of a kernel with the image data. The convolution process includes determining cumulative sums for pixels in the image portion, particularly focusing on differences between cumulative sums of adjacent pixels in a second row of the image portion. These differences are summed to produce a value for a target pixel. The resulting sum is then multiplied by the element values of the kernel region to complete the convolution operation. This approach minimizes redundant calculations by reusing intermediate results, thereby improving computational efficiency. The technique is particularly useful in applications requiring real-time image processing, such as computer vision, medical imaging, and autonomous systems, where reducing computational overhead is critical. By optimizing the convolution process, the method enables faster and more efficient image analysis without sacrificing accuracy. The invention builds on prior methods by introducing a more streamlined calculation process that leverages cumulative sums and kernel element values to enhance performance.
15. A method according to claim 13 , wherein: convolving the kernel with the image portion further includes at least determining a sum of the difference between the cumulative sums for the two pixels of the second row and the difference between the cumulative sums for the two pixels of the second row, to determine the value for the target pixel; and normalising said differences sum as a function of a number of elements in the kernel.
This invention relates to image processing, specifically an optimized method for applying a convolution kernel to an image portion. The problem addressed is the computational inefficiency of traditional convolution operations, particularly in real-time or resource-constrained applications. The method involves processing an image portion by convolving a kernel with the image data. The convolution process includes calculating cumulative sums for pixels in a second row of the image portion. The method then determines a sum of the differences between these cumulative sums for two pixels in the second row. This difference sum is used to compute a value for a target pixel in the image. The resulting difference sum is normalized based on the number of elements in the kernel to produce the final output value. The technique leverages cumulative sums to reduce the number of arithmetic operations, improving computational efficiency. By focusing on differences between cumulative sums, the method minimizes redundant calculations, making it suitable for applications requiring fast or low-power processing, such as edge devices or embedded systems. The normalization step ensures the output remains consistent regardless of kernel size. This approach enhances performance while maintaining accuracy in convolution-based image processing tasks.
16. An image processing application for an image processing device, the image processing application being arranged to perform the steps of claim 1 .
An image processing application for an image processing device processes digital images to enhance visual quality. The application includes a noise reduction module that analyzes pixel data to identify and reduce noise artifacts, such as graininess or speckling, while preserving image details. A color correction module adjusts color balance, saturation, and hue to improve naturalness and consistency. An edge enhancement module sharpens edges and fine details to increase clarity. The application also includes a dynamic range adjustment module that optimizes contrast and brightness for better visibility in both bright and dark regions. Additionally, a compression module reduces file size without significant quality loss, making the processed images suitable for storage or transmission. The application may operate in real-time or batch mode, depending on the processing requirements. It is designed to work with various image formats and can be integrated into cameras, smartphones, or standalone software. The system improves image quality for professional and consumer use, addressing issues like poor lighting, sensor noise, and compression artifacts.
17. The application of claim 16 being arranged to provide a plurality of blurring options for an acquired image including: at least one custom 2-dimensional kernel based blur; and one or more of a motion blur, zoom blur or a spin blur.
This invention relates to image processing techniques for applying various blurring effects to acquired images. The technology addresses the need for flexible and customizable blurring options in digital imaging systems, allowing users to enhance visual effects or reduce noise while maintaining control over the appearance of the final image. The system provides a plurality of blurring options, including at least one custom 2-dimensional kernel-based blur, which allows users to define their own blur patterns by specifying a matrix of weights. This enables precise control over the blur effect, such as adjusting the intensity, direction, or shape of the blur. Additionally, the system offers predefined blur effects, including motion blur, zoom blur, and spin blur. Motion blur simulates the effect of movement during image capture, creating streaks or trails that follow the direction of motion. Zoom blur mimics the effect of adjusting the camera lens during exposure, producing a radial distortion that emphasizes the center of the image. Spin blur creates a circular or rotational effect, often used to convey a sense of rotation or dynamic motion. The system is designed to be integrated into imaging devices or software applications, providing users with a versatile toolset for post-processing images. The combination of customizable kernel-based blurring and predefined motion effects allows for both artistic expression and technical adjustments, catering to a wide range of applications in photography, video editing, and digital art.
18. An image processing system, comprising: memory configured to store at least a portion of a digital image comprising rows of pixels extending in two perpendicular directions over a 2D field; at least one processor configured to perform operations including: for each pixel in at least a row of pixels within said image portion, calculating a cumulative sum for said pixel as a sum of a value of said pixel and a value representing a sum of values of preceding pixels in the same row of pixels, wherein the values used for calculating the cumulative sum for said pixel include only values of pixels in the same row of pixels; and convolving a region of a kernel with the image portion, wherein a target pixel in the image is aligned with a specified element of the kernel region, including at least determining a difference between the cumulative sum for said pixel and a cumulative sum for at least one other pixel of the row of pixels, to determine a value for the target pixel.
This invention relates to an image processing system designed to efficiently perform convolution operations on digital images. The system addresses the computational challenges associated with traditional convolution methods, which often require repeated calculations for overlapping pixel regions, leading to inefficiencies. The system includes memory storing a digital image divided into rows of pixels arranged in two perpendicular directions over a 2D field. A processor performs operations to optimize convolution by calculating cumulative sums for pixels in a row. For each pixel in a row, the processor computes a cumulative sum as the sum of the pixel's value and the cumulative sum of all preceding pixels in the same row. This ensures that only pixel values from the same row are used, avoiding cross-row dependencies. During convolution, a kernel region is applied to the image portion. A target pixel is aligned with a specified element of the kernel region. The system determines the target pixel's value by calculating the difference between the cumulative sum of the target pixel and the cumulative sum of at least one other pixel in the row. This approach leverages precomputed cumulative sums to accelerate convolution, reducing redundant calculations and improving processing speed. The system is particularly useful in applications requiring real-time image processing, such as computer vision, medical imaging, and autonomous systems, where efficient convolution operations are critical.
19. The image processing system of claim 18 , the operations further comprising: providing a plurality of pre-defined kernels for selection by a user, receiving from a user a selection of one of said pre-defined kernels, and using a region of said selected kernel as the kernel region for processing said image portion.
The invention relates to image processing systems designed to enhance or modify digital images through kernel-based operations. A common challenge in image processing is applying filters or transformations that adapt to specific image regions, requiring precise kernel selection and application. The system addresses this by offering a set of pre-defined kernels, each representing a different filter or transformation pattern. Users can select a kernel from this set, and the system then applies a region of the chosen kernel to process a corresponding portion of the image. This approach allows for flexible and user-customizable image processing, enabling adjustments to brightness, contrast, edge detection, or other effects based on the selected kernel. The system dynamically maps the selected kernel region to the image portion, ensuring accurate and localized modifications. This method improves efficiency by reducing the need for manual kernel design while providing tailored processing for different image areas. The invention is particularly useful in applications requiring adaptive filtering, such as medical imaging, photography, or computer vision tasks.
20. The image processing system according to claim 18 , wherein: the row is a first row; the operations further comprise, for each pixel in at least a second row of pixels within said image portion, calculating a cumulative sum for said pixel as a sum of a value of said pixel and a value representing a sum of values of preceding pixel values pixels in the second row of said image portion pixels; and convolving the region of the kernel with the image portion further includes at least determining a difference between the cumulative sum for said pixel and a cumulative sum for at least one other pixel of the second row of pixels, to determine the value for the target pixel.
The invention relates to an image processing system designed to efficiently perform convolution operations on image data. Convolution is a fundamental operation in image processing, often used for tasks such as blurring, sharpening, edge detection, and feature extraction. Traditional convolution methods require repeated multiplication and addition operations, which can be computationally intensive, especially for large kernels or high-resolution images. This invention addresses the inefficiency by leveraging cumulative sums to accelerate the convolution process. The system processes an image by dividing it into portions and applying a convolution kernel to each portion. For each pixel in a first row of an image portion, the system calculates a cumulative sum representing the sum of the pixel's value and the sum of preceding pixel values in the row. Similarly, for each pixel in a second row, the system computes a cumulative sum. During convolution, the system determines the value for a target pixel by calculating the difference between the cumulative sum of the target pixel and the cumulative sum of another pixel in the same row. This approach reduces the number of arithmetic operations needed, improving computational efficiency. The method can be applied iteratively across multiple rows and columns to fully convolve the image portion with the kernel. The invention is particularly useful in real-time image processing applications where speed and efficiency are critical.
Unknown
November 10, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.